Prediction models for COVID-19 disease outcomes.

Area of Science:

• Infectious Diseases
• Genomics
• Machine Learning

Background:

• Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has caused millions of deaths and long-term health issues.
• COVID-19 presents a wide spectrum of illness severity, from asymptomatic cases to fatalities.
• Effective management requires accurate prediction of individual patient outcomes.

Purpose of the Study:

• To develop personalized risk assessment models for predicting clinical outcomes in COVID-19 patients.
• To identify key predictors of hospitalization, intensive care unit (ICU) admission, and long COVID.
• To inform targeted interventions and enhance patient monitoring.

Main Methods:

• Retrospective cross-sectional study of 4450 individuals in Missouri with sequenced SARS-CoV-2 samples.
• Integration of virus genomic data with clinical history, disease course, and urban-rural classification.
• Development of machine learning models to predict hospitalization, ICU admission, and long COVID.

Main Results:

• Predictors for hospitalization included immunosuppression, cardiovascular disease, older age, specific symptoms, rural residence, and viral genetic markers.
• ICU admission was associated with acute respiratory distress syndrome, ventilation, bacterial co-infection, rural residence, and non-wild type SARS-CoV-2 variants.
• Long COVID was linked to hospital admission, ventilation, and female sex.

Conclusions:

• Developed risk assessment models can identify COVID-19 patients needing enhanced monitoring or early interventions.
• Integrating virus, host, and environmental factors is crucial for accurate patient outcome prediction.
• These models provide a platform for personalized medicine adaptable to other infectious diseases.